Following evaluation of the options concerned, Internode has taken the decision to invest in building a new international point of presence (PoP) site.

The new site will be in Tokyo, Japan.

This will be implemented as a 'breakout' of our existing capacity purchases from Australia via Japan to the USA on the AJC cable system.

Its not as simple as it might sound to do this - because getting from the coastal landing sites for AJC and other cable systems through to Tokyo is a bit of a challenge - its about 100 km's between the two, so the gigabit fibre patch leads need a little help :)

But... we're doing it - and indeed we are already well advanced on making it happen.

I won't provide a commit date on it because there are several components to making this work and we're trying to get them all done before we declare victory. We are expecting, however, that it should all be done within a month or two - again, we're already well advanced on much of it.

We'll be putting in an initial capacity breakout of the order of 622 Megabits per second, Sydney to Tokyo, which should be adequate for a start :)

What this will do:

- Add one more country to our network map

- Dramatically lower the latency of data flows between Australia and Japan (and onward to various other Asian destinations)

- Improve our redundancy and options in terms of gaining further transit connectivity to the world over time, and improve our global availability of access to the Internet for the same reason.

- Link us to a new market with new opportunities in it, for business and residential customers:

Business customers may find this boosts their ability to use Internode to reach Asian destinations efficiently; Residential customers should find this will improve performance for applications such as VoIP and online gaming into significant parts of Asia.

- Increase our global peering (we'll peer with Japanese Internet providers and content backbones such as Google Japan, etc etc).

We have a similar opportunity to establish a breakout into New Zealand, for largely similar reasons, and that option is currently being evaluated as well. Note that an NZ breakout is not a commitment at this time - only the Japan one is committed at this point.

There will be some media releases about this in due course, but I thought you might want to hear about it first, here :)

This is not a cheap exercise (to put it mildly) - but in all sorts of ways, its something we just had to do.

Regards,
Simon Hackett

You heard it first on the Washing Machine.

Nice work, the desire to expand the pretty map was clearly the driving force.

Good work, Simon. thanks for the heads up.

given the strength of the asian economies and the booms as china and india move out of the 3rd world economy it is probably a long sighted move.

Thats awesome! :D

Simon Hackett writes...

But... we're doing it - and indeed we are already well advanced on making it happen.

Another reason to add to the list on why Internode is going from strength to strength :)
Good news Simon!
Quick question: Would this reduce traffic latency to broader places like Europe?

Shafto writes...

Quick question: Would this reduce traffic latency to broader places like Europe?

Probably not. Long story short, most transit from Asia to Europe runs via the USA.

There are cable paths to Europe via the middle east, but there are various negatives about them (including the distance not being much different in the end), and such that sheer economics tends to drive most traffic the other way around the ball.

Shafto writes...

Quick question: Would this reduce traffic latency to broader places like Europe?

Not really - Europe via Northern Asia is very close to Europe via the US.

Until a new cable from Perth to Singapore or India comes up there's not going to be a great way of reducing it significantly.

Hey. This would sound great to get to in NZ

~Luke

Ahh I see, that makes sense.
Thanks for the replies Simon & Matthew.
I got a second opinion! ;)

Shame there is still no official support for Western Europe. :(
The NZ pop sounds good, lots lower ping for there game servers. :P

Binary Smells writes...

The NZ pop sounds good, lots lower ping for there game servers. :P

I'm interested in any high ping NZ servers so we can ensure any connectivity we do there works well.

Matthew Moyle-Croft writes...

I'm interested in any high ping NZ servers so we can ensure any connectivity we do there works well.

Fix = Find somone over there on dial up to host a server :D

Binary Smells writes...

Fix = Find somone over there on dial up to host a server :D

??

I'm trying to be serious - if there are specific servers or providers with higher latency than just AUS to NZ would imply, then I'd like to know what they are so that we can see if any connectivity will fix that.

Matthew Moyle-Croft writes...

if there are specific servers or providers with higher latency than just AUS to NZ would imply

Might be worth getting Kingsley to query that on the games forum (if you haven't already). I'm sure there will be a bunch of Unreal Tournament players just itching to give you some good targets (I seem to recall that had been a bit of a thorn on occasion). ;)

NZ Internet is **** Connections are **** too. We Need a new Company to provide a PoP

Simon Hackett writes...

We have a similar opportunity to establish a breakout into New Zealand, for largely similar reasons, and that option is currently being evaluated as well.

Awesome!

Note that an NZ breakout is not a commitment at this time - only the Japan one is committed at this point.

Please make this happen :)

Call me Bob writes...

Awesome!

Please make this happen :)

What's wrong with connectivity into NZ as it stands right now? I don't think that I've ever had any real problems with it.

Its rubbish. Slow. they say as fast as your line handles. Most places maxing it at 50kb to 300KB

If u lucky

Simon Hackett writes...

- Dramatically lower the latency of data flows between Australia and Japan (and onward to various other Asian destinations)


Awesome!
(last time I was in Japan, traceroute showed that I was being reached via a rather indirect path).

Awesome news.

Very happy to see this news/info :)

Lukeintosh writes...

Its rubbish. Slow. they say as fast as your line handles. Most places maxing it at 50kb to 300KB

Well, while we have no plans to become an ISP in NZ (thats a whole 'nuther kettle of prawns), we'd be more than happy to sell rocking global transit connections to ISPs currently in NZ that wanted one :)

bradp writes...

What's wrong with connectivity into NZ as it stands right now? I don't think that I've ever had any real problems with it.

Mostly it's ok. Though there are times when NZ traffic seems to take a route via the US.

Maybe when NZ wins another ANZAC test we may consider an ISP moving into NZ. :P

Simon Hackett writes...

This will be implemented as a 'breakout' of our existing capacity purchases from Australia via Japan to the USA on the AJC cable system

Can you explain a bit more on that.

damo81 writes...

Can you explain a bit more on that.

Internode's AJC links already travel through Japan, but from there they've currently only got connectivity to the US.

IE - Even though it's already landing in Japan, it's essentially just acting as a big link direct to the US, because that's the only place it's got to go.

Installing a new PoP in Japan means that once data hits that point - instead of just routing it automatically through to the US - they'll actually be able to route it onwards to other locations as well [IE - the rest of Asia].

niiiice :D

Call me Bob writes...

Mostly it's ok. Though there are times when NZ traffic seems to take a route via the US.

I've experienced this too, but I don't have any specific examples on hand.

Simon Hackett writes...

Well, while we have no plans to become an ISP in NZ (thats a whole 'nuther kettle of prawns)

Don't you mean 'a whole 'nuther kittle of fush' ??

:)

Lukeintosh writes...

Its rubbish. Slow. they say as fast as your line handles. Most places maxing it at 50kb to 300KB

not want ing to burst burst your bubble, but evur thurt ur just on sa shurt line? :-P

Matthew Moyle-Croft writes...

Until a new cable from Perth to Singapore or India comes up there's not going to be a great way of reducing it significantly.

I would have thought India would be more Viable, given that *MOST* of the Connectivity currently goes through Alexandrea (SP) where they had the cable cut recently, which disconnected 75% of India's connectivity.

Or how about India, via Singapore ?

Brian White writes...

Or how about India, via Singapore ?

Part of the issue is getting capacity going west from Australia.

Matthew, here's a New Zealand server i'ed love to see fixed, it's a Quake 3 server.

gamers.net.nz ultra freeze lpb
60.234.73.177:27965

Routing to it is fine, but we get routed back via the USA.
Internode have previously tried to fix this, but to no avail, as it was up to telstraclear (i think) to play ball, they didn't want to.

Michael Kratz writes...

Part of the issue is getting capacity going west from Australia.

I already thought there was a cable going from Perth to Singapore (SEA-ME-WE 3)?

Fantastic news. Anyone here who has setup intranet VPNs to Asia and witnessed the mayhem from last years earthquake will know what I mean. Quite often the best latency to China was when packets went through Japan (this was on Optus).

Businesses will pay a premium to make life easier dealing with Asia.

Mr B writes...

gamers.net.nz ultra freeze lpb
60.234.73.177:27965

Routing to it is fine, but we get routed back via the USA.

Seems perfectly fine to me.

Tracing the path to 60.234.73.177 on TCP port 27965, 30 hops max
1 lns4.mel6.internode.on.net (150.101.212.22) 14.264 ms 14.647 ms 14.778 ms
2 gi1-3.cor1.mel6.internode.on.net (150.101.212.158) 12.979 ms 15.349 ms 12.308 ms
3 gi5-2.cor3.mel4.internode.on.net (150.101.212.136) 25.853 ms 25.984 ms 27.534 ms
4 pos2-1.bdr1.syd6.internode.on.net ­ (150.101.120.122) 27.475 ms 25.705 ms 25.338 ms
5 Gi14-0-3-503.gw2.syd1.asianetcom.n­ et (150.101.197.34) 26.543 ms 25.406 ms 26.292 ms
6 po6-0.cr2.syd1.asianetcom.net (202.147.40.138) 26.291 ms 25.737 ms 25.570 ms
7 so-0-0-0-0.gw4.akl1.asianetcom.net­ (202.147.55.253) 50.439 ms 48.657 ms 53.399 ms
8 Orcon-Internet.gw4.AKL1.asianetcom­ .net (203.192.166.70) 50.688 ms 50.172 ms 50.929 ms
9 gi-5-8.dsw1.nct.orcon.net.nz (60.234.9.24) 51.422 ms 50.179 ms 50.913 ms
10 60.234.73.177 [closed] 51.928 ms 50.261 ms 50.732 ms

webby writes...

I already thought there was a cable going from Perth to Singapore (SEA-ME-WE 3)?

forum-replies.cfm?t=9104...479927#r14479927

Not enough capacity for us.

webby writes...

I already thought there was a cable going from Perth to Singapore (SEA-ME-WE 3)?

Its very old - and very full. No more room at the inn there (at any price).

There are some efforts underway to raise capital to build a new cable from Perth up into Asia. But its not happened yet.

Moar conectvitivy = good news (and well done, too!)

Time for some pwning of the Japanese over XBL :P

Download speed increases in general from jp hosts, will also be welcome

Mr B writes...

Routing to it is fine, but we get routed back via the USA.
Internode have previously tried to fix this, but to no avail, as it was up to telstraclear (i think) to play ball, they didn't want to.

Yes, that should be a good example of something that would become a non-issue if we built a breakout in New Zealand in our own right.

The reason the return traffic is heading via the USA is that BGP routing is not only complex, but its also quite dumb in some ways (ways that tend to increase its resilience, but which can be a problem at boundary points like this). The point is that the nearest Internode router on the return path, in the view of that system... is the one in the USA.

("Nearest" in BGP, long story short, tends to mean 'lowest number of router hops' - not shortest physical distance - which isn't measured or stored in BGP)

Obviously if we fielded a router in NZ, then that'd be the closest, and it would therefore tend to 'attract' the return packets properly back into itself.

Simon

Binary Smells writes...

The NZ pop sounds good, lots lower ping for there game servers.

Since when did NZ get more than a few cans connected by string? They only got leckky a couple of years ago. :-)

CypherOz writes...

Since when did NZ get more than a few cans connected by string? They only got leckky a couple of years ago. :-)

[OT]Umm, actually until internode rolled out ADSL2, their speeds were better than anything on offer here (Thanks Telstra). OT]

Brian White writes...

I would have thought India would be more Viable, given that *MOST* of the Connectivity currently goes through Alexandrea (SP) where they had the cable cut recently, which disconnected 75% of India's connectivity.

Or how about India, via Singapore ?

While it's all very fine suggesting "nice" places to go if there isn't any actual demand to pay for it then you've just burned a whole lot of money for no additional revenue. At the moment pricing of significant capacity into India would take your breath away, and it's cheaper to just bypass it onto Europe.

The broken cables also illustrate another problem in that if you want a reliable service then you're going to have to buy capacity on two systems to your ultimate destination. When crossing the Pacific you can do this just with SCCN since it has two paths on its figure of 8 system but that doesn't apply when heading to Europe. Of course both SMW4 and Flag failed simultaneously recently so if you had used both of them for your protection you would have still lost out anyway. SMW3's reliability is terrible compared to the other two but it managed to avoid the submarine back hoe that took out SMW4 and Flag.

in NZ would you plug into www.citylink.co.nz - or www.nzix.net ?

maybe on nodephone, dialing from AU to NZ could be a 18c "untimed national" call. :-)
How about a NZ DID?

MIAH

miah writes...

in NZ would you plug into www.citylink.co.nz - or www.nzix.net ?

We'd most likely peer at the Auckland exchange in line with our general peering strategy. We're still exploring what options exist and what we want to achieve in NZ, so beyond that I can't comment.

Matthew Moyle-Croft writes...

I'm interested in any high ping NZ servers so we can ensure any connectivity we do there works well.

This traceroute is good enough for me

Target Name: gpstore.com.au
IP: 202.89.36.243
Date/Time: 3/20/2008 10:53:36 AM

2 29 ms 29 ms 32 ms lns10.syd7.internode.on.net [150.101.197.27]
3 30 ms 29 ms 29 ms vl14.cor2.syd7.internode.on.net [150.101.120.83]
4 30 ms 30 ms 29 ms po3.bdr1.syd6.internode.on.net [150.101.199.232]
5 30 ms 30 ms 30 ms Gi14-0-3-503.gw2.syd1.asianetcom.n­ et [150.101.197.34]
6 31 ms 31 ms 30 ms po0-1.cr2.syd1.asianetcom.net [202.147.40.125]
7 53 ms 53 ms 53 ms so-0-0-0-0.gw4.akl1.asianetcom.net­ [202.147.55.253]
8 54 ms 54 ms 52 ms gi9-3.gw1.akl1.asianetcom.net [202.147.55.209]
9 55 ms 56 ms 55 ms MaximumInternet.gw1.akl1.asianetco­ m.net [202.147.41.66]
10 55 ms 54 ms 56 ms gi5-1.hsa1.alb.maxnet.net.nz [123.100.64.131]
11 58 ms 56 ms 55 ms gpstore.com.au [202.89.36.243]

Simon Hackett writes...

we'll peer with Japanese Internet providers and content backbones such as Google Japan

Google Japan?... how does that work out? i thought that you JUST peer with Google in the USA.

Another GSR12000 going in there also?

Great work once again!
Internode are getting bigger and better than ever :>

Cant wait to see Internode in 10+ yrs.
Rulers :> haha

Google Japan? I thought Yahoo/YahooBB was the only content provider in Japan.

This investment is mostly for the benefit of gamers right? What kind of latency can we expect playing games on servers in Japan, Korea and China? I think I can already get about 200ms now. Palying with the American and European community is usually too laggy and time difference is an issue too. It will also be good to bet decent download speeds from Japanese servers.

Any chance of a breakout from the AJC north through the Bering Strait, under the Arctic ice-cap, then down to Iceland, and into Europe?

Just hire an ex-soviet submarine, load it up with cable fed out of the missile tubes, rename it the SS Internode and you're away.

If you could get that done by June, I can read the Guardian website with slightly less latency.

As an added benefit, you'd get lower latency to the East coast.

And here's the kicker - open a "green" server colo in Iceland with natural cooling AND geothermal power!. No CO2 emmissions!

Mr Bubbles writes...

This investment is mostly for the benefit of gamers right?

No ... it's going to [positively] affect anything that's latency dependent, and that covers a lot more than just gaming.

VoIP is one big example, as well as people who have VPNs etc into Asian countries.

What kind of latency can we expect playing games on servers in Japan, Korea and China?

Try some traceroutes from another ISPs network to see [roughly] what you can expect.

einstein.dontexist.net/austrace

I think I can already get about 200ms now.

Not with Internode you won't be!

It will also be good to bet decent download speeds from Japanese servers.

There shouldn't be any problems with download speeds as they are right now. They're an example of something that's not adversely affected by higher latency.

LittleRascal writes...

Any chance of a breakout from the AJC north through the Bering Strait, under the Arctic ice-cap, then down to Iceland, and into Europe?

Better get-in-line.

Internode will undoubtedly consider it more important to run backhaul from Fawkner to Coburg exchange so that I can have ADSL2+. Sorry, ahem, so that all the good townsfolk connected to the Fawkner exchange can have ADSL2+. ;-)

And it's only 4 kms or so, unlike your undersea cable. And it doesn't require a submarine.

This is terrific news. Will improve streaming video where the remote server uses domestic (Japan) buffer sizes no end.

Now can Internode lead the market by making the link work faster than the speed of light so we can get sub 80mS ping times to Japan?

It's on the to-do list isn't it Simon?

Kevmeister writes...

Sorry, ahem, so that all the good townsfolk connected to the Fawkner exchange can have ADSL2+. ;-)

How many Polar Bears are there in Fawkner? At a guess, not many. Glaciers? I don't think so.

Never underestimate the importance of polar bears in undersea cable network design.

Mr Bubbles writes...

Google Japan? I thought Yahoo/YahooBB was the only content provider in Japan.

Err no. There's heaps of them in Japan.

I tell you what!

Great news XD

As for the SS Internode... Wishful Thinking LOL

I just had another great idea for the SS Internode.

A cable to Cape Town via Antarctica.

We go south from Hobart to McMurdo, then across the ice suspending the cable on indestructable Hill's Hoists, then north to Cape Town.

It would finally give everyone a reason to put capacity into Tasmania.

LittleRascal writes...

I just had another great idea for the SS Internode.

A cable to Cape Town via Antarctica.

We go south from Hobart to McMurdo, then across the ice suspending the cable on indestructable Hill's Hoists, then north to Cape Town.

A fine plan in all its glory but sadly the cost of backhaul from Tas to Vic scuttles the whole idea in one fell swoop.

Proudly brought to you (at exorbitant prices) by Telstra. Keep on smiling.

Just tell me this please....why was Simon Hackett not given the job as boss of Telstra?

Pretty sure this guy is the most innovative, and proactive business man in the telecommunications industry. He also is a straight shooter (yet still diplomatic), and (to the best of my knowledge) is an honest man with integrity.

He has surrounded himself with a knowledgable, and highly capable team, and together I can see no better team around.

Love reading what you have to say, and also what you and Internode do, Simon.

Congrats.

i like ya way of thinking ....dam clothes lines are indestructable

audifred writes...

Just tell me this please....why was Simon Hackett not given the job as boss of Telstra?

<snip>

You just answered your own question :)

Adam

Woohoo.

My Japanese MMOs will now play better!

good work, keep on expanding!

Hai Nyuu writes...

My Japanese MMOs will now play better!

I'm interested in any IPs you have of MMO servers.

does wow have any asian servers? or all in America?

battlenet connections should definately benefit

Hobbsy writes...

does wow have any asian servers?

Yes, in Korea and China.

or all in America?

Nope ... they've also got plenty of servers around Europe as well.

wowwiki.com/Realms_List

Matthew Moyle-Croft writes...

I'm interested in any IPs you have of MMO servers.

I play Gundam Tactics Online. I'll try and see if I can dig up the IP address for you tonight, though it's not a game that relies too much on latency.

So WoW performance should pick up a bit if people are willing to move realms. I assume that pings to Asia are around half that to the US?

Transfers to WoW Korea, China or Taiwan are impossible. For China and Korea, you might even need a Social Security number of some kind as when I tried to apply for Korean RO it asked me for it.

Hobbsy writes...

I assume that pings to Asia are around half that to the US?

LOL ... not quite.

Japan still pings about 150ms return from Sydney.

well it's still a small improvement.

audifred writes...

Just tell me this please....why was Simon Hackett not given the job as boss of Telstra?

They probably offered him the job, but he turned it down as he knows that some things are just not worth fixing :P

Hobbsy writes...

well it's still a small improvement.

It's a large improvement. At the moment it goes up past Japan, to the US then back to Japan. So probably around 15 to 20,000km further than it has to I believe + a couple of extra hops.

Speed of light is 299,792,458 meters per second. So to travel 20,000,000 meters would take 67ms. Now you incur this penalty there and back, so in terms of ping times it will equate to approximately 130ms.

For example, a Team Fortress 2 server from Japan (www.game-monitor.com/sea...=tf2&location=JP):

Pinged it from my home machine in Adelaide:
C:\Documents and Settings\Admin>ping 121.112.159.131

Pinging 121.112.159.131 with 32 bytes of data:

Reply from 121.112.159.131: bytes=32 time=326ms TTL=48
Reply from 121.112.159.131: bytes=32 time=327ms TTL=48
Reply from 121.112.159.131: bytes=32 time=327ms TTL=48
Reply from 121.112.159.131: bytes=32 time=326ms TTL=48

Ping statistics for 121.112.159.131:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss)
Approximate round trip times in milli-seconds:
Minimum = 326ms, Maximum = 327ms, Average = 326ms

And now pinged from my web server (in the US):
C:\Documents and Settings\Administrator>ping 121.112.159.131

Pinging 121.112.159.131 with 32 bytes of data:

Reply from 121.112.159.131: bytes=32 time=173ms TTL=47
Reply from 121.112.159.131: bytes=32 time=176ms TTL=47
Reply from 121.112.159.131: bytes=32 time=176ms TTL=47
Reply from 121.112.159.131: bytes=32 time=173ms TTL=47

Ping statistics for 121.112.159.131:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 173ms, Maximum = 176ms, Average = 174ms

Now pinging that US server from my box:
C:\Documents and Settings\Admin>ping

Pinging *** * with 32 bytes of data:

Reply from ***: bytes=32 time=220ms TTL=114
Reply from ***: bytes=32 time=220ms TTL=114
Reply from ***: bytes=32 time=220ms TTL=114
Reply from ***: bytes=32 time=220ms TTL=114

Ping statistics for ***:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 220ms, Maximum = 220ms, Average = 220ms

So it'd take approximately 390ms to go from my box, to the US and back to Japan which is an extra 170ms. And note that the US box is in the middle of the US, where as Internode would connect to the West coast so that explains the extra 40ms from the calculated 130ms saving.

Also note that Australia is closer to Japan than the US is to Japan (slightly). So sub 150ms pings should be possible, especially from Sydney.

If I wanted to be more accurate I could have done trace routes and picked where they cross in LA to calculate the actual difference.

Cling Wrap writes...

Speed of light is 299,792,458 meters per second.

In a vacuum ...

Light only travels ~2/3 that speed through fiber optics.

So to travel 20,000,000 meters would take 67ms. Now you incur this penalty there and back, so in terms of ping times it will equate to approximately 130ms.

US <-> Japan takes around about 110ms round trip coast to coast.

Cling Wrap writes...

Also note that Australia is closer to Japan than the US is to Japan (slightly). So sub 150ms pings should be possible, especially from Sydney.


This is currently with iinet:

Pinging 121.112.159.131 with 32 bytes of data:

Reply from 121.112.159.131: bytes=32 time=136ms TTL=47
Reply from 121.112.159.131: bytes=32 time=136ms TTL=47
Reply from 121.112.159.131: bytes=32 time=136ms TTL=47
Reply from 121.112.159.131: bytes=32 time=136ms TTL=47

Ping statistics for 121.112.159.131:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 136ms, Maximum = 136ms, Average = 136ms

I am very :) at the moment :)

Cling Wrap writes...

Speed of light is 299,792,458 meters per second.

... in a vacuum. In glass it's about two thirds of that (more or less; varies slightly depending on the optical density of the glass)

- mark

Cling Wrap writes...

Also note that Australia is closer to Japan than the US is to Japan (slightly). So sub 150ms pings should be possible, especially from Sydney.

Yeah - cable station to cable station the latency is about 112ms.

So, ADSL2+ Low latency in Sydney to a host in Tokyo should be in the order of 130-140ms.

Can someone enlighten me as to what content we are currently pulling from Japan/Asia which doesn't fit the below categories:

1. connections to Asian offices (branches / HQ's etc)
2. Gaming Servers
3. People visiting web sites in their motherland etc

Just curious. I understand that Google has a fair amount of content delivered out of Japan, so is peering with them in Japan going to be better than bringing it across from the US?

Maybe im to ocker to know otherwise - just not sure what kind of interesting content comes from Japan (or Asia). What am I missing out on?

bradp writes...

Light only travels ~2/3 that speed through fiber optics.

I also made another mistake by doubling the distance and then doubling the latency when I should have just doubled one. Although coincidently my combination of errors ended up being pretty accurate in terms of real world performance.

Simon Hackett writes...

decision to invest in building a new international point of presence (PoP) site.

Well done Internode, keep up the excellent work!!!

©£ONeST@® writes...

This is currently with iinet:

This is what I get with 'Node on the Ulladulla NSW exchange.

>ping 121.112.159.131

Pinging 121.112.159.131 with 32 bytes of data:
Reply from 121.112.159.131: bytes=32 time=336ms TTL=49
Reply from 121.112.159.131: bytes=32 time=335ms TTL=49
Reply from 121.112.159.131: bytes=32 time=334ms TTL=49
Reply from 121.112.159.131: bytes=32 time=334ms TTL=49

Ping statistics for 121.112.159.131:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 334ms, Maximum = 336ms, Average = 334ms

>tracert 121.112.159.131

Tracing route to i121-112-159-131.s11.a023.ap.plala­ .or.jp [121.112.159.131]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms ipcop.localdomain [192.168.0.1]
2 * 29 ms 30 ms lns10.syd7.internode.on.net [150.101.197.27]
3 29 ms 29 ms 29 ms vl14.cor2.syd7.internode.on.net [150.101.120.83]

4 186 ms 185 ms 186 ms gi0-3.bdr1.syd7.internode.on.net [150.101.197.46
]
5 185 ms 185 ms 185 ms pos2-0.bdr1.sjc2.internode.on.net ­ [203.16.213.41]
6 187 ms 187 ms 188 ms ge-3-7.r02.snjsca04.us.bb.gin.ntt.­ net [129.250.11.97]
7 213 ms 214 ms 214 ms ae-2.r21.snjsca04.us.bb.gin.ntt.ne­ t [129.250.4.9]
8 313 ms 336 ms 314 ms as-0.r21.tokyjp01.jp.bb.gin.ntt.ne­ t [129.250.5.82]
9 285 ms 286 ms 286 ms 129.250.11.54
10 287 ms 288 ms 287 ms 210.145.252.185
11 283 ms 284 ms 283 ms 221.184.4.10
12 323 ms 324 ms 323 ms 218.43.251.254
13 283 ms 283 ms 284 ms i210-225-225-222.s99.a049.ap.plala­ .or.jp [210.225.225.222]
14 328 ms 330 ms 329 ms 210.225.224.188
15 331 ms 349 ms 339 ms 58.95.254.138
16 334 ms 337 ms 335 ms i121-112-159-131.s11.a023.ap.plala­ .or.jp [121.112.159.131]

Trace complete.

FTTH Test Case writes...

2. Gaming Servers

Thats a big one for me.
With telstra it was a good sub 200 ms.
With Internode its a good 300 ms.

It will be nice to play with my friends in Japan again.

FTTH Test Case writes...

What am I missing out on?

I think it's important to understand that we have a large Asian population in Australia, who want to look at content in their first/native language. So you personally may not be missing out on a lot, it'll help our customers that do want faster access to that content.

audifred writes...

why was Simon Hackett not given the job as boss of Telstra?

Because then he'd be forced into making as much money as possible and not the most innovative, and proactive business man in the telecommunications industry.

Being CEO of Telstra is not as easy as some may think.

No doubt Simon is happy with where he is now.

Mark Newton writes...

... in a vacuum. In glass it's about two thirds of that (more or less; varies slightly depending on the optical density of the glass)

Damn that cursed Luminiferous aether!

Will you be selling private IP to japan for corp customers?

kemza writes...

Will you be selling private IP to japan for corp customers?

Sure - we've been focusing on building the POP itself, but certainly one of the reasons for doing so is to enable MPLS VPN services for corporate customers (we do this out of our US POPs already).

Would this improve transfer speeds from manufacturers in Taiwan and China or is that problem largely there own connections not being big enough?
EG near all computer hardware makers. The largest and most common that comes to mind is www.realtek.com.tw which normally has terrible speeds.

Will the new PoP site be in Equinix,Tokyo? or where

And will you be connecting to other networks in there?

damo81 writes...

Will the new PoP site be in Equinix,Tokyo?

Yes.

And will you be connecting to other networks in there?

Thats kinda the point :)

Simon Hackett writes...

Thats kinda the point :)

I knew that :)

Simon Hackett writes...

There are some efforts underway to raise capital to build a new cable from Perth up into Asia. But its not happened yet.

That'd be nice to see. I kinda like redundancy, and not being forced to route everything through the US.

As I understand it, we pay to connect to the US, the UK pays to connect to the US, and when we want to connect to the UK then we both have to pay while the US does not. Maybe I'm being petty or misunderstanding the system though :)

GregAlex writes...

That'd be nice to see. I kinda like redundancy, and not being forced to route everything through the US.

I kinda do, too. Thats one of the reasons for the investment in a PoP in Japan.

As I understand it, we pay to connect to the US, the UK pays to connect to the US, and when we want to connect to the UK then we both have to pay while the US does not. Maybe I'm being petty or misunderstanding the system though :)

Long story short, everyone else pays to reach the USA, in terms of international circuit costs.

At a deep level, the global Internet is subsidised, in the USA, by the rest of the world, as a result. Its one of the reason for better (Internet) deals there, although the gap in value terms progressively continues to close (as the relative cost of international circuits lowers over time).

Are you still going to keep the AJC links via japan to the USA

damo81 writes...

Are you still going to keep the AJC links via japan to the USA

of course, we want to keep the redundancy of having links to the US via two separate cable systems.

Wont purchasing more bandwidth cost more money meaning higher costs and more price increases

Brad Amm writes...

Wont purchasing more bandwidth cost more money meaning higher costs and more price increases

No ...

To get there right now they've got to pay for capacity into the US, and then from there to shunt the data back to Asia.

Every mbit that uses this new [direct] link, doesn't have to make that double trip, and is also 1 mbit less that's going to be taking up capacity on the direct US links.

Simon Hackett writes...

We'll be putting in an initial capacity breakout of the order of 622 Megabits per second, Sydney to Tokyo, which should be adequate for a start :)

so a 622Mb/s SYD - LAX1/SJC2 becomes
622Mb/s SYD - JPN1 and 622Mb/s JPN1 - LAX1/SJC2
Will the redundancy to Japan will be by LAX1/SJC2 and SXC?

p.s. I like the open and transparentness of the Internode Network, it gives me confidence in the network you are building, and helped me choose NODE for my ADSL,
SO, when will internode customers be able to see the cacti graphs
( like monitor.waia.asn.au/cact...d=143&rra_id=all )
for the complete network? i'm a graph watcher!!!

miah writes...

SO, when will internode customers be able to see the cacti graphs

They've got something even better.

nodemap.internode.on.net

miah writes...

SO, when will internode customers be able to see the cacti graphs


Sorry, but you won't. That data is commercially sensitive, and hence, won't be made public :)

Michael Kratz writes...

Sorry, but you won't. That data is commercially sensitive, and hence, won't be made public :)

You don't have to hide this, its a selling point. like your network map, its also commercial info, but you show that! and i'm the type that respect YOU for showing it - not many other ISPs do show, but i USE internode because you are open to discuss your network, ( - average people- who don't CARE about their networks use DODO or BIGPOND etc)

I'm not just any 'joe blow' member of the public, i'm a NODE customer, i have a commercial relationship with NODE, i'd look at signing a NDA. Easy to make it available to only NODE IP ranges or somthing!

do you mind WAIX publishing their cacti? maybe PIPE should publish their cacti ALSO !

Are the savings going to be passed onto the customer, which was brought up by Simon only a short time ago.

miah writes...

do you mind WAIX publishing their cacti? maybe PIPE should publish their cacti ALSO !

They already do.

You just need a login to the PIPE DB to see this - so you'll either be a carrier or a company with it's own IP space which you are advertising to the IX. In other words, the graphs are there, but they are not open to the general public.

But what does it really matter what the graphs and figures look like? You're buying a service from an ISP - and how it is bolted together is irrelevant unless you can suggest there is an issue causing some impact. Just the same as you don't expect graphs of utilisation of other utilities such as your street water supply, neighbourhood KWh usage from a power company, or delivery details of gas for your local service station....

Get on with enjoying the Internet service now and in the future as that is what you are buying, not wasting time looking for and thinking about things like utilisation graphs :->

miah writes...

You don't have to hide this, its a selling point. like your network map, its also commercial info, but you show that!

You missed the word "sensitive". Obviously that specific information is too sensitive to be given to the general public/competitors. The same that at work we are not permitted to give information to the public relating to actual stock levels at our store, we're allowed to tell them if we only have "a couple" of "plently" but not actual numbers.

Do you mean the network map?

damo81 writes...

Do you mean the network map?

No he means actual data flow through the network map, not what it's capable of.

sorry about the OT comments, back to the topic of
> Internode to establish PoP in Japan

Woohoo, this is great news!

Will this provide us with any advantages with Youtube streaming for example? or is youtube streaming from the US anyway?

webby writes...

Will this provide us with any advantages with Youtube streaming for example? or is youtube streaming from the US anyway?

Won't change Youtube.

But hopefully will speed up sites like gigabyte.com.tw and other similar hardware vendors.

rocket writes...

But hopefully will speed up sites like gigabyte.com.tw and other similar hardware vendors.

I already asked about this and got no answer in regards to it

bits writes...

I already asked about this and got no answer in regards to it

I guess we will have to wait and see.

rocket writes...

But hopefully will speed up sites like gigabyte.com.tw and other similar hardware vendors.

I don't really see how?

The new PoP will definitely improve latency/redundancy to those places, but latency really isn't something that has a big impact on actual download speeds.

The 'problem' with a lot of Asia, is that they've got HUGE amounts of bandwidth within a particular country, but often very poor connectivity between neighbouring countries.

rocket writes...

But hopefully will speed up sites like gigabyte.com.tw and other similar hardware vendors.

now thats something that needs attention.
its always slow for me.

rocket writes...

But hopefully will speed up sites like gigabyte.com.tw and other similar hardware vendors.

In gigabyte.com.tw's case you might be in luck. Their ISP appears to be Chunghwa Telecom and tracerouting from their traceroute web page shows a link between Taiwan & Japan to NTT. If you traceroute to IIJ or KDDI (two other big Japanese ISPs) they bounce traffic off the US west coast.

Just because these countries are "close" within North Asia doesn't mean there is connectivity between them in Asia. Most ISPs see more value in connecting to the Internet in the US and so bouncing traffic off the US west coast is the norm rather than an exception.

Of course this also assumes that both sides select the same path. Chunghwa Telecom might prefer to use IIJ over NTT.

There's quite a few cables coming out of Japan: www1.alcatel-lucent.com/...World_Map_LR.pdf

Impressive! I didn't realise (from wiki, so take it with a grain of salt) that the under seas cables are only 70mm in diameter, tiny!!

MrWhite writes...

under seas cables are only 70mm in diameter, tiny!!

That's the armoured cable. The deep-sea stuff is more like 15 - 20mm.

- mark

Mark Newton writes...

That's the armoured cable

I'm armoured, grrrrr!! Take that sharks!! (Ok, back to water for me) Thanks for the clarification Mark.

Simon

So does this mean that when I try to stream local Japan TV (from my rellos house who have unlimited mega speed fibre with YahooBB) with a Slingbox (also maybe Sony location free player or PS3 to PSP and even ORB or similar) to my pc here I will get a better Kbps rate that the abysmal rate from Japan as it is now?

theslydog writes...

to my pc here I will get a better Kbps rate that the abysmal rate from Japan as it is now?

Well, that entirely depends on if it removes some choke point that exists at the moment.

(If you could whim me or email me the IP you're going to in Japan then I'd be interested in having a look).

I just spent three hours using remote desktop to a server in Tokyo trying to solve an issue we are having with sending and processing large files (300MB+) there.

A very painful experience - all error logs in Japanese and large latency.

Good to hear that one of these issues will be resolved soon!

one writes...

I just spent three hours using remote desktop to a server in Tokyo

Any chance you could whim/email me an IP address (or range)?

Wonderful, now i can finally get steaming videos from my Japanese friends to work :)

Not to mention stream some of the jpop stations, and モーニング娘 live concerts... drools

ありがとハケトさん！
，

Matthew Moyle-Croft writes...

I'm interested in any IPs you have of MMO servers.

goin.perfectworld.com.my

I doubt it will be worth it tho. Up to japan and then back to malaysia is probably 300ms+
I suspect packet loss will also be a problem for anyone not in .my

Then again if Internode hosted a server themselves.....
One can hope.

Rika-san writes...

Wonderful, now i can finally get steaming videos from my Japanese friends to work :)

Not to mention stream some of the jpop stations, and モーニング娘 live concerts... drools

ありがとハケトさん！
，

in english... i might look into it. all i can read is a 1 and a 2 these days.

vladesch writes...

goin.perfectworld.com.my

Ping statistics for 210.48.156.136:
Packets: Sent = 32, Received = 32, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 331ms, Maximum = 396ms, Average = 347ms

This is what I got from a Bigpond Extreme connection in Melbourne, if it helps to see things from the other side...

vladesch writes...

I doubt it will be worth it tho. Up to japan and then back to malaysia is probably 300ms+

I suspect it'll be less than 300ms.

Kingsley writes...

in english... i might look into it. all i can read is a 1 and a 2 these days.

He said now he can stream live morning musume concerts (15 - 21 year old girls who sing in miniskirts aimed for pervy guys etc)

blog.drecom.jp/caramelda.../img20070507.jpg

and he said thank you to haketo san ? mr hackett i guess.

Dope fish writes...

blog.drecom.jp/caramelda.../img20070507.jpg

Hmm I must visit Japan one day. :p

Things like this make me want to churn from tpg over to Internode. The price increase does not worry be if i know how you are spending it. I.E these sorts of things i'm totally for and wouldn't mind paying extra to help this be achieved. Nice one Internode. Can't wait for ADSL2+ in my area!

Maybe when this is completed you'll have another customer :)

goin.perfectworld.com.my

Ping statistics for 210.48.156.136:
Packets: Sent = 32, Received = 32, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 331ms, Maximum = 396ms, Average = 347ms

Fowl writes...

This is what I got from a Bigpond Extreme connection in Melbourne, if it helps to see things from the other side...

It would be interesting to see a traceroute of that.

vladesch writes...

It would be interesting to see a traceroute of that.

www.telstra.net/cgi-bin/trace

Or if you want to try from some other networks as well

einstein.dontexist.net/austrace

My Bad, Already Answered. (Just got in to china, to much going on to remmeber Whirlpool posts :(

The reason why i ask is that, we are going to be needing to link up locations in Singapore, Shinzen, Shanghai, Hong Kong as a VPN back to Australia. Was thinking of making them go through a carrier in Japan to haul it back to Sydney. We will need this around Sep so hopefully Internode has there POP finished around then :)

kemza writes...

Will Internode sell private IP to Japan?

We won't have a product completely sorted until after the POP is live, but the intention is to be able to.

Have you forgotten the answer to that question already?

forum-replies.cfm?t=9408...942326#r14942326

kemza writes...

The reason why i ask is that, we are going to be needing to link up locations in Singapore, Shinzen, Shanghai, Hong Kong as a VPN back to Australia. Was thinking of making them go through a carrier in Japan to haul it back to Sydney. We will need this around Sep so hopefully Internode has there POP finished around then :)

You probably need to check out how ISPs such as Singtel and Pacnet get from Asia to Australia. Hong Kong based ISPs might head to Guam and down to Oz while Singtel has capacity on SMW3 into Perth.

Simon Hackett writes...

We are expecting, however, that it should all be done within a month or two - again, we're already well advanced on much of it.

I make that 5 weeks or so.

(note: it's not fully integrated into our routing, so most people won't quite get the benefit until we do that, but it's live and working)

MMC
At NRT1

5:51pm.tsunami.~% traceroute www.ntt.com
traceroute to www.ntt.com (61.208.134.143), 30 hops max, 38 byte packets
1 gw (x.x.x.x) 1.511 ms 0.812 ms 0.769 ms
2 lnsx.adlx.internode.on.net (x.x.x.x) 22.481 ms 22.175 ms 22.790 ms
3 vl29-cor1.adl6.internode.on.net (203.16.212.6) 23.327 ms 25.587 ms 23.070 ms
4 po3.bdr1.adl6.internode.on.net (203.16.212.146) 159.397 ms 160.514 ms 160.522 ms
5 pos4-0.bdr1.syd7.internode.on.net ­ (203.16.212.21) 160.656 ms 160.405 ms 160.358 ms
6 pos2-0-0.bdr1.nrt1.internode.on.ne­ t (203.16.211.34) 163.075 ms 160.730 ms 160.752 ms
7 203.105.73.137 (203.105.73.137) 242.958 ms 160.886 ms 160.753 ms
8 203.105.73.81 (203.105.73.81) 160.708 ms 160.777 ms 160.724 ms
9 xe-6-2-0.a21.tokyjp01.jp.ra.gin.nt­ t.net (61.120.147.205) 161.670 ms 160.469 ms 161.029 ms
10 ae-0-1.a21.tokyjp01.jp.ra.gin.ntt.­ net (61.213.161.218) 161.379 ms 161.420 ms 160.965 ms
11 210.145.252.185 (210.145.252.185) 161.453 ms 160.891 ms 160.967 ms
12 60.37.54.201 (60.37.54.201) 161.432 ms 160.865 ms 161.171 ms
13 60.37.54.166 (60.37.54.166) 161.780 ms 161.872 ms 161.693 ms
14 222.146.39.62 (222.146.39.62) 162.419 ms 161.369 ms 161.945 ms
15 221.184.31.154 (221.184.31.154) 162.102 ms 162.085 ms 161.478 ms
16 * *

let me be the first to say...

Well done !!

feel sorry for the poor guy doing your network map,
you guys make that almost a full time job :)

Great work guys... Traceroute to canon.jp

4 159 ms 159 ms 159 ms pos6-0.bdr1.syd7.internode.on.net ­ [203.16.212.25]
5 159 ms 159 ms 159 ms pos2-0-0.bdr1.nrt1.internode.on.ne­ t [203.16.211.34]
6 160 ms 159 ms 159 ms 203.105.73.137
7 160 ms 159 ms 159 ms 203.105.73.81
8 160 ms 159 ms 160 ms xe-5-3-0.a21.tokyjp01.jp.ra.gin.nt­ t.net [61.213.169.105]
9 160 ms 159 ms 160 ms ae-1.a10.tokyjp01.jp.ra.gin.ntt.ne­ t [61.120.147.222]
10 160 ms 160 ms 160 ms fa-2-3-4.a10.tokyjp01.jp.ra.gin.nt­ t.net [61.213.145.206]
11 161 ms 161 ms 160 ms 202.228.142.9
12 * ^C

Jay Binks writes...

feel sorry for the poor guy doing your network map,

That would be me :)

you guys make that almost a full time job :)

The diagram with the Japan POP is already drawn and ready; it'll go up when everything is fully in production and all the appropriate connections are active.

How about a POP in Mainland China or something? That would be mad.

Matthew Moyle-Croft writes...

15 221.184.31.154 (221.184.31.154) 162.102 ms 162.085 ms 161.478 ms

To get an idea of how much of an improvement this is I ran a traceroute to the same general destination earlier today which came up with:

11 comxp-sw126.nttcom.co.jp (163.137.2.6) 308.744 ms 309.256 ms 308.627 ms

So that's about half the round trip time.

Have a look at other destinations in Asia for more network goodness For instance I've seen sites in Hong Kong routing this way this evening :-)

jsl

John Lindsay writes...

Have a look at other destinations in Asia for more network goodness For instance I've seen sites in Hong Kong routing this way this evening :-)

I've seen Singapore and Taiwan too. Great work guys :)

Edit: some Thailand and Malaysia too. NTT sure have a big network :)

Are all the Peers setup and connected to the Japan PoP yet or is that a progressive thing?

waagu writes...

Are all the Peers setup and connected to the Japan PoP yet or is that a progressive thing?

NTT transit is up. One peering exchange is connected, but no peers yet. The other one will be turned up tomorrow then I can started doing some peering.

i pretty clueless to this so i will ask anyway.

is there an actual cable that runs on the ocean floor?

Yes. AJC.

- mark

Mark Newton writes...

Yes. AJC.

oh wow, kinda vulnerable to attacks. cant be to many global links from Australia to abroad.

oh well a story for another day :)

rrrohan writes...

oh wow, kinda vulnerable to attacks.

Yeah, you can see the massive queues of enemies of the state just lining up to attack it if you look closely at a zoomed-in view on Google Earth.

Oh wait, no you can't. What was I thinking.

cant be to many global links from Australia to abroad.

AJC and SCCN is just about it for now. We buy capacity on both.

- mark

does this mean better speed from Korean/Taiwan based http/ftp sites?

Esp those motherboard manufacturers

Mark Newton writes...

Yeah, you can see the massive queues of enemies of the state just lining up to attack it if you look closely at a zoomed-in view on Google Earth.

lol can just imagine a terrorist screaming a bubbly "ALLAH!!" as they swim down to cable.

one more question. if there is more that one route out of Australia is it just luck of the draw which route it takes or is there some intelligence to it?

aLcAtRoN writes...

Esp those motherboard manufacturers

my god Asus are shocking for that.

120ms (ish) to Internode Japan pop.

<3 Internode

It looks like the Japan PoP is active :)

Target Name: hk.yahoo.com
IP: 202.43.220.99
Date/Time: 4/23/2008 8:16:15 AM

2 * 18 ms 19 ms 19 ms 17 ms lns10.syd7.internode.on.net [150.101.197.27]
3 18 ms 18 ms 18 ms 17 ms 17 ms vl14.cor2.syd7.internode.on.net [150.101.120.83]
4 134 ms 134 ms 132 ms 132 ms 133 ms gi0-3.bdr1.syd7.internode.on.net [150.101.197.46]
5 132 ms 134 ms 170 ms 140 ms 133 ms pos2-0-0.bdr1.nrt1.internode.on.ne­ t [203.16.211.34]
6 133 ms 133 ms 136 ms 132 ms 133 ms [203.105.73.137]
7 134 ms 133 ms 136 ms 133 ms 133 ms [203.105.73.81]
8 132 ms 133 ms 134 ms 133 ms 132 ms xe-5-3-0.a21.tokyjp01.jp.ra.gin.nt­ t.net [61.213.169.105]
9 132 ms 133 ms 152 ms 134 ms 132 ms xe-2-1-0.r20.tokyjp01.jp.bb.gin.nt­ t.net [61.213.162.101]
10 237 ms 238 ms 237 ms 237 ms 237 ms ge-0.sprint.tokyjp01.jp.bb.gin.ntt­ .net [129.250.9.46]
11 281 ms 279 ms 278 ms 277 ms 277 ms sl-bb20-hk-14-3.sprintlink.net [203.222.33.10]
12 276 ms 273 ms 272 ms 272 ms 272 ms sl-gw10-hk-9-0-0.sprintlink.net [203.222.38.106]
13 256 ms 254 ms 255 ms 255 ms 255 ms sla-yahoo2-1-0.sprintlink.net [203.222.39.74]
14 281 ms 281 ms 280 ms 280 ms 281 ms w.hp.vip.hki.yahoo.com [202.43.220.99]

rrrohan writes...

is it just luck of the draw which route it takes or is there some intelligence to it?

BGP is our friend and hundreds of thousands of dollars worth of Cisco routers work out the best (usually shortest) path to send traffic to. We live in hope that the people at the other end apply similar amounts of care in sending packets back.

We have gone to some considerable effort to make this exercise worthwhile and in general Asia will be much better for Internode customers now and much better than most other ISPs in Australia.

jsl

Have Japanese relatives in Japan and since we are using nodephone it was good anyway to ring them but now it will be even better and then theres skype.....

Good work John and Simon.

So will the rest of the Japan PoP will be fully active today?

thargs writes...

in Japan and since we are using nodephone it was good anyway to ring them but now it will be even better

that's a thought- does Internode have voice interconnects in overseas pops or is the voice side of things handled differently?

Awesome, just awesome. Keep pushing the envelope Internode :)

Wow, just picked out a random counter-strike server in Japan..

1 0ms 0ms 0ms crus.gateway [192.168.1.254]
2 5ms 5ms 5ms lns10.syd7.internode.on.net [150.101.197.27]
3 5ms 5ms 5ms vl14.cor2.syd7.internode.on.net [150.101.120.83]
4 120ms 120ms 120ms gi0-3.bdr1.syd7.internode.on.net [150.101.197.46]
5 120ms 120ms 120ms pos2-0-0.bdr1.nrt1.internode.on.ne­ t [203.16.211.34]
6 120ms 120ms 120ms 203.105.73.137
7 120ms 120ms 120ms 203.105.73.81
8 121ms 120ms 120ms xe-5-3-0.a20.tokyjp01.jp.ra.gin.nt­ t.net [61.213.169.101]
9 120ms 120ms 120ms xe-7-1.a14.tokyjp01.jp.ra.gin.ntt.­ net [203.105.72.90]
10 121ms 120ms 120ms xe-1-2.a14.tokyjp01.jp.ra.gin.ntt.­ net [61.213.161.6]
11 120ms 120ms 120ms tkybi5.asahi-net.or.jp [202.224.32.85]
12 127ms 127ms 127ms ngynirb.asahi-net.or.jp [124.155.64.9]
13 163ms 134ms 129ms ngynig1.asahi-net.or.jp [203.212.37.65]
14 131ms 130ms 131ms x049195.ppp.asahi-net.or.jp [122.249.49.195]

Nice work node! :D

Looks promising to me :)

Browsing Singapore sites now significantly faster. Thanks, Internode.

crusj0r writes...

Wow, just picked out a random counter-strike server in Japan..

Asahi are one of the people I'm hoping to peer with. This will mean we can probably cut 5 or so hops off that.

Small Factoid: This is Internode's first POP that is dualstack IPv4/IPv6 from the beginning. Our transit with NTT is dual stack and we see 25% of the IPv6 routes via Japan.

Ironically neither of the peering fabrics we're connecting to in Japan do IPv6 as dual stack yet where as PIPE has 3 in Australia.

Bryn writes...

or is the voice side of things handled differently?

Generally differently but that's because voice termination into a market is almost always cheaper from outside that market. For instance long distance voice is insanely expensive from Tokyo to destinations inside Japan. However in many markets international voice is usually insanely cheap. We could save money by "refiling" our calls to Australia in the USA. Of course it would negatively impact quality but it would be super cheap.

jsl

Looks awesome, when are we going to see an updated network map?

not_using_telstra writes...

Looks awesome, when are we going to see an updated network map?

/forum-replies.cfm?t=940808&p=8#r141

Fuzion writes...

How about a POP in Mainland China or something? That would be mad.

One in the EU would be nice too :p

not_using_telstra writes...

One in the EU would be nice too :p

I'm sure one day we'll get to Europe :-)

Domo arigato Internode?

E C H O writes...

Domo arigato Internode?

iie?

Yay for IPv6.


3 vl67.cor1.adl6.internode.on.net (2001:44b8:8060:8000::1) 20.360 ms
4 gig0-3.bdr1.adl6.internode.on.net ­ (2001:44b8:8060:3::1) 19.965 ms
5 pos3-1.bdr1.syd7.internode.on.net ­ (2001:44b8:8060:1::2) 42.964 ms
6 pos2-0-0.bdr1.nrt1.internode.on.ne­ t (2001:44b8:f0a0:1::1) 158.113 ms
7 fa-2-2.r01.tokyjp01.jp.b6.gin.ntt.­ net (2001:218:2000:5000::d) 157.625 ms
8 xe-1-4.a16.tokyjp01.jp.ra.gin.ntt.­ net (2001:218:2000:2000::89) 158.391 ms
9 xe-5-3-0.a20.tokyjp01.jp.ra.gin.nt­ t.net (2001:218:2000:2000::125) 158.410 ms
10 ae-0-1.a20.tokyjp01.jp.ra.gin.ntt.­ net (2001:218:2000:5000::4a) 157.076 ms

ありがとう　ございました！

What he said :)

haha yeah!

Excellent!
Saw my pings to SingNet fall by 150ms yesterday at 5:30pm.
Now if we could get some IPv6 goodness through Node too. :)

2 ge-1-0-0.bb1.a.syd.aarnet.net.au
(2001:388:1:5001:212:1eff:fe8e:51f­ e) 29.836 ms 30.589 ms 31.474 ms
3 ge-0-0-0.bb1.b.syd.aarnet.net.au
(2001:388:1:d:212:1eff:fe90:7000) ­ 32.293 ms 33.078 ms 33.732 ms
4 so-2-1-0.bb1.a.lax.aarnet.net.au
(2001:388:1:30::2) 193.345 ms 193.831 ms 194.909 ms
5 p4-1-1-0.r03.lsanca03.us.bb.gin.nt­ t.net
(2001:418:4000:5000::9) 195.705 ms 196.101 ms 197.013 ms
6 xe-3-3-0.r20.lsanca03.us.bb.gin.nt­ t.net
(2001:418:0:2000::2d9) 197.762 ms 201.657 ms 186.594 ms
7 as-2.r20.osakjp01.jp.bb.gin.ntt.ne­ t
(2001:218:0:2000::ae) 337.157 ms 301.592 ms 302.359 ms
8 ae-4.r20.tokyjp01.jp.bb.gin.ntt.ne­ t
(2001:218:0:2000::d9) 325.452 ms 309.377 ms 327.424 ms
9 xe-0-0-0.a20.tokyjp01.jp.ra.gin.nt­ t.net
(2001:218:0:6000::1a) 334.204 ms 312.881 ms 314.987 ms

Simon Hackett writes...

What he said :)

hahaha

Matthew Moyle-Croft writes...

Asahi are one of the people I'm hoping to peer with. This will mean we can probably cut 5 or so hops off that.

Love ya work - perfect example of why i stay with and recommend internode.

Matthew Moyle-Croft writes...

Asahi are one of the people I'm hoping to peer with. This will mean we can probably cut 5 or so hops off that.

Is JPIX anywhere on the radar as well?

www.jpix.ad.jp/en/user/user.html

bradp writes...

Is JPIX anywhere on the radar as well?

It is, according to the Internode page on peeringdb.

bradp writes...

Is JPIX anywhere on the radar as well?

Later today, in all likelihood. MMC is there right now.

- mark

Mark Newton writes...

MMC is there right now

More up to date than the Pipe blog. I love it :)

Mark Newton writes...

Later today, in all likelihood. MMC is there right now.

LOL ... speak of the devil ... it looks like it's actually just come on right now!

Mark Newton writes...

Later today, in all likelihood. MMC is there right now.

Now actually - just turned up.

% traceroute www.jpix.co.jp
traceroute to www.jpix.co.jp (210.171.225.81), 30 hops max, 40 byte packets
1 vl15.cor3.adl2.internode.on.net (203.16.215.62) 2.394 ms 0.687 ms 0.482 ms
2 gi0-3.bdr1.adl6.internode.on.net (150.101.134.51) 272.668 ms 207.092 ms 207.170 ms
3 pos4-3.bdr1.syd7.internode.on.net ­ (203.16.212.37) 137.222 ms 137.226 ms 137.239 ms
4 pos2-0-0.bdr1.nrt1.internode.on.ne­ t (203.16.211.34) 137.347 ms 137.473 ms 137.533 ms
5 AS7527.ix.jpix.ad.jp (210.171.224.3) 138.180 ms 172.737 ms 148.703 ms
6 ge-0-1-0-ext-gw1.jpix.ad.jp (210.171.226.234) 138.110 ms 158.262 ms 138.906 ms
7 eth3-1-bb-fw-12.jpix.ad.jp (210.171.226.100) 138.270 ms 138.310 ms 138.166 ms
8 web1.jpix.ad.jp (210.171.225.81) 138.320 ms 141.192 ms 138.292 ms

Matthew Moyle-Croft writes...

Now actually - just turned up.

I won! :p

Looks good though.

Started at ~295ms before the new link come online.
Down to ~235ms with the new link, but no peering. [16 hops]
Down to ~150ms with the direct peering. [8 hops]

bradp writes...

I won! :p

Really?

bdr1.nrt1#ping www.jpix.co.jp
Translating "www.jpix.co.jp"...domain server (192.231.203.132) (192.231.203.3)

Translating "www.jpix.co.jp"...domain server (192.231.203.132) [OK]

Type escape sequence to abort.
Sending 5, 100-byte ICMP Echos to 210.171.225.81, timeout is 2 seconds:
!!!!!
Success rate is 100 percent (5/5), round-trip min/avg/max = 1/1/4 ms

Matthew Moyle-Croft writes...

Really?

No ... I mean that I beat you to making the post! :p

Anyway ... is there any sort of story with PCCW? Traffic to them still seems to be heading via the US.

Are there any plans/chance of peering with them somewhere as well? They're at JPIX by the looks of it.

bradp writes...

Anyway ... is there any story with PCCW? Traffic to them still seems to be heading via the US.

This will get resolved in the near to medium future.

Don't know if its a direct result of the changes but I can now receive the (electric car) stream below without rebuffering. (A 1Mbit stream).

www.mitsubishi-motors.tv...x_5tateyama.html

/Rant: Pity the Gov with its 'big ideas' doesn't foster 'green' car manufacture like Thailand is doing.

Matthew Moyle-Croft writes...

Now actually - just turned up.

Oooh, have you tried whale meat yet? I heard its a bit expensive now :-)

Oi³ writes...

Oooh, have you tried whale meat yet? I heard its a bit expensive now :-)

No - I doubt I will either.

All these Asian sites and I cant read Asian. :(
I like the lower ping to the Asian game servers tho. :D

Well done Node, next pop NZ?

MIAH

Like the press release said, we'd like to.

I'm making enquiries, look at it that way.

- mark

Mark Newton writes...

Like the press release said, we'd like to.

I'm making enquiries, look at it that way.

So when they lose "yet another" ANZAC rugby test we can all DDoS there servers. :P

Yep. And when they win they can DDoS ours. :-)

- mark

looking-glass.aussiehq.net.au

Tracing the route to yahoo.co.jp (124.83.139.192)

1 gi2-51.cor1.cbr1.internode.on.net ­ (150.101.160.90) 16 msec 0 msec 0 msec
2 pos1-1.bdr1.syd7.internode.on.net ­ (150.101.120.118) [AS 4739] 120 msec 120 msec 116 msec
3 pos2-0-0.bdr1.nrt1.internode.on.ne­ t (203.16.211.34) [AS 4739] 120 msec 120 msec 116 msec
4 203.105.73.137 [AS 2914] 120 msec 116 msec 120 msec
5 203.105.73.77 [AS 2914] 116 msec 120 msec 368 msec
6 xe-3-0-0.a21.osakjp01.jp.ra.gin.nt­ t.net (61.213.162.206) [AS 2914] 128 msec 128 msec 128 msec
7 61.200.82.42 [AS 2914] 200 msec 200 msec 248 msec
8 124.83.128.33 [AS 24572] 200 msec 200 msec 200 msec
9 124.83.128.26 [AS 24572] 204 msec 204 msec 204 msec
10 124.83.128.42 [AS 24572] 208 msec 208 msec 208 msec
11 * * *
12 * * *

Traces from hong kong and tokyo seem to be coming back through the USA and down to us

is this because everything is not connected at the PoP
eg:peers and other IX's

Tracing route to www.internode.on.net [203.16.214.27]
over a maximum of 30 hops:

1 1 ms <1 ms <1 ms 192.168.1.1
2 7 ms 9 ms 8 ms 10.44.255.254
3 7 ms 7 ms 8 ms 192.168.51.2
4 6 ms 7 ms 8 ms 192.168.51.6
5 8 ms 9 ms 8 ms 210-200-111-50.cr.static.apol.com.­ tw [210.200.111.50]
6 8 ms 11 ms 8 ms 210-58-88-34.cm.static.apol.com.tw­ [210.58.88.34]
7 11 ms 8 ms 11 ms 210-200-105-222.ebix.net.tw [210.200.105.222]
8 11 ms 11 ms 12 ms 62.216.146.117
9 39 ms 41 ms 45 ms so-4-3-0.0.cjr04.tok002.flagtel.co­ m [62.216.128.14]
10 149 ms 158 ms 150 ms so-1-1-0.0.cjr02.lax002.flagtel.co­ m [62.216.128.34]
11 * * * Request timed out.
12 249 ms 254 ms 249 ms pos6-0.bdr1.syd6.internode.on.net ­ [203.16.213.157]
13 255 ms 249 ms 252 ms pos3-2.bdr2.adl2.internode.on.net ­ [203.16.212.141]
14 253 ms 249 ms 250 ms gi1-17.cor1.adl6.internode.on.net ­ [150.101.134.52]
15 249 ms 251 ms 248 ms pubweb.internode.on.net [203.16.214.27]

This is a trace from Taiwan. Previously we we're getting around 320ms to our site in Melbourne, now it's around 250ms. My guess from the trace is that it's still going to the US huh? But the latency is a great improvement and we are seeing about 20% increase on ftp uploads to Melb.

damo81 writes...

Traces from hong kong and tokyo seem to be coming back through the USA and down to us

The PoP is physically installed. The process of sorting out optimal routing (to the greatest extent possible) has only just begun.

Even when thats done, there may well be some specific instances where routing will be asymmetric - that isn't at all unusual in multi-country international network situations of this sort.

As you can see, there is already an improvement evident now - and hopefully it'll keep getting better as we keep tuning what we can (its not 100% under our control - return data paths are a function of where the other guys' routers think the optimal path lies - something we can influence but not control)

Regards,
Simon

OnederBoy writes...

My guess from the trace is that it's still going to the US huh?

Sort of. In your case, as per the post above this one, what is likely is that the traffic in one direction is now taking a more optimal path, and the return traffic is not (yet) doing so.

As you've noted, even that first step has yielded visible improvement.

Simon

Yeah definately no complaints here Simon, if anything thanks for the continued improvement on the Node network.

Great work guys, it just gets better and better. Cheers!

OnederBoy writes...

Great work guys, it just gets better and better. Cheers!

Just think in 10 years how good its going to be :) :)

Michael Kratz writes...

The diagram with the Japan POP is already drawn and ready; it'll go up when everything is fully in production and all the appropriate connections are active.

Not sure if this is a Michael question or a John/Simon question, but are Node doing a Dual/redundant POP scenario in JP similar to what they have done on the west coast of the US (ie: SJ and LA POPs)???

(Apologies if this has been discussed elswhere in the thread)...

Rossco

Rossco writes...

Not sure if this is a Michael question or a John/Simon question, but are Node doing a Dual/redundant POP scenario in JP similar to what they have done on the west coast of the US (ie: SJ and LA POPs)???

Not at this stage. There will only be a single POP in Japan for now, it's possible that may be re-evaluated at a later stage.

Note though, that the AJC capacity to the U.S. doesn't go 'through' this POP per-se, only through the cable stations at Shima and Maruyama. So the AJC capacity to and from the US that goes via Japan is not dependent on the operational status of this POP.

Bluemoon writes...

Now if we could get some IPv6 goodness through Node too. :)

I'll sure it'll be here when it's good and ready.

Tracing the route to xe-0-0-0.a20.tokyjp01.jp.ra.gin.nt­ t.net (2001:218:0:6000::1A)

1 gi0-267.cor1.cbr1.internode.on.net­ (2001:44B8:1010:60::4739:1) 0 msec 0 msec 0 msec
2 po1-1.bdr1.syd7.internode.on.net (2001:44B8:9010:2::7) 4 msec 4 msec 4 msec
3 pos2-0-0.bdr1.nrt1.internode.on.ne­ t (2001:44B8:F0A0:1::1) 116 msec 116 msec 116 msec
4 fa-2-2.r01.tokyjp01.jp.b6.gin.ntt.­ net (2001:218:2000:5000::D) 128 msec 120 msec 128 msec
5 xe-4-4.a15.tokyjp01.jp.ra.gin.ntt.­ net (2001:218:2000:2000::85) 120 msec 120 msec 120 msec
6 xe-0-0-0.a20.tokyjp01.jp.ra.gin.nt­ t.net (2001:218:0:6000::1A) 120 msec 120 msec 120 msec

/me enjoys dual stack IPv6.

:)

Simon Hackett writes...

there may well be some specific instances where routing will be asymmetric - that isn't at all unusual in multi-country international network situations of this sort.

Thanks Simon..got it :)

forget my post before stupid question

Tracing route to www.internode.on.net [203.16.214.27]
over a maximum of 30 hops:

1 1 ms <1 ms <1 ms 192.168.1.1
2 9 ms 7 ms 9 ms 10.44.255.254
3 25 ms 13 ms 7 ms 192.168.51.2
4 7 ms 8 ms 7 ms 192.168.51.6
5 9 ms 21 ms 9 ms 210-200-111-50.cr.static.apol.com.­ tw [210.200.111.50]
6 11 ms 7 ms 7 ms 210-58-88-42.cm.static.apol.com.tw­ [210.58.88.42]
7 10 ms 9 ms 10 ms 210-200-105-222.ebix.net.tw [210.200.105.222]
8 12 ms 17 ms 11 ms 62.216.146.117
9 41 ms 44 ms 43 ms so-4-3-0.0.cjr04.tok002.flagtel.co­ m [62.216.128.14]
10 * * * Request timed out.
11 179 ms 179 ms 177 ms pos3-3.bdr1.syd7.internode.on.net ­ [203.16.211.33]
12 176 ms 179 ms 178 ms pos2-0.bdr1.adl6.internode.on.net ­ [203.16.212.182]
13 178 ms 184 ms 182 ms gi1-17.cor3.adl2.internode.on.net ­ [150.101.134.50]
14 179 ms 176 ms 179 ms pubweb.internode.on.net [203.16.214.27]

Just reviving this thread. This is a trace from our site in Taiwan to Internode in Adelaide. Our Node is in Melbourne which is even less by about 5ms.

Previously uploading files from Melbourne to Taiwan topped out at around 700-800kbps. Now we're looking at around 1200-1300kbps. Melbourne Annex M syncs around 1950.

Pings dropped from 320ms (on low ping game mode) to 245ms after Japan POP came online. Now it's 170ish ms so I guess the routing is all optimized.

I just remember Simon saying a few months back about once Japan coming online, he 'defies' anyone to beat them on speed.

Well I don't have anything to compare, considering I'm paying the same money, this improvement is AWESOME!

Thanks Internode, you guys really rock...

Can someone on Node do a tracert to sohu.com and qq.com for me please. Thanks :)

OnederBoy writes...

Now it's 170ish ms so I guess the routing is all optimized.

Some of the routing side of things just takes time as it involves negotiating with other organisations. We've not finished yet, things will continue to improve over the next few months for futher locations.

170ms is probably as good as it'll get from Taiwan - it's 112ms from SYD to Tokyo and then a bit to Taiwan from there.

I get 120/130ms to japan and 160/170ms to west coast usa.
No more yank servers for me I can tell you. :D

I heard using IPv6 will lower pings even more, is this true?
And that IPv6 will stop ddos altogether as it does not use some packet?

SnipSnap! writes...

I heard using IPv6 will lower pings even more, is this true?

No, it'll raise them ever so slightly as routers have to work harder (longer addresses use more processing power/hardware to match).

Most of the cause of latency is due to distance and the speed of light through glass.

And that IPv6 will stop ddos altogether as it does not use some packet?

No, it won't stop it.

IPv6 ain't magic. It's just this protocol you know. (with apologies to HHGTTG).

Mikey- writes...

Can someone on Node do a tracert to sohu.com and qq.com for me please.

Tracing route to sohu.com [61.135.181.176]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.1.254
2 10 ms 11 ms 11 ms lns1.bne1.internode.on.net [150.101.180.16]
3 10 ms 10 ms 11 ms gi2-3.cor2.bne1.internode.on.net [150.101.180.115]
4 179 ms 178 ms 178 ms pos4-2.bdr1.syd6.internode.on.net ­ [203.16.212.13]
5 187 ms 187 ms 186 ms pos2-0.bdr1.lax1.internode.on.net ­ [203.16.213.198]
6 187 ms 186 ms 187 ms 12.118.42.21
7 189 ms 189 ms 189 ms tbr1.la2ca.ip.att.net [12.122.82.142]
8 187 ms 187 ms 188 ms gar3.lsrca.ip.att.net [12.122.104.17]
9 443 ms 436 ms 426 ms 12.116.103.10
10 * 423 ms 420 ms 219.158.3.133
11 418 ms 423 ms 423 ms 202.96.12.182
12 420 ms 422 ms 421 ms 61.148.3.34
13 441 ms 433 ms 439 ms 202.106.48.66
14 422 ms 423 ms 423 ms 61.135.181.176

Trace complete.

Tracing route to qq.com [219.133.40.91]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.1.254
2 11 ms 11 ms 11 ms lns1.bne1.internode.on.net [150.101.180.16]
3 11 ms 10 ms 10 ms gi2-3.cor2.bne1.internode.on.net [150.101.180.115]
4 203 ms 203 ms 203 ms pos4-2.bdr1.syd6.internode.on.net ­ [203.16.212.13]
5 202 ms 202 ms 202 ms pos4-0.bdr1.lax1.internode.on.net ­ [203.16.213.189]
6 211 ms 211 ms 212 ms 12.118.42.21
7 223 ms 223 ms 222 ms tbr1.la2ca.ip.att.net [12.122.82.142]
8 224 ms 223 ms 224 ms cr1.la2ca.ip.att.net [12.122.19.153]
9 248 ms 248 ms 247 ms cr1.sffca.ip.att.net [12.122.3.121]
10 215 ms 215 ms 215 ms tbr1.sffca.ip.att.net [12.122.19.10]
11 246 ms 245 ms 244 ms ggr3.sffca.ip.att.net [12.122.86.25]
12 214 ms 214 ms 215 ms 192.205.35.86
13 588 ms 585 ms 572 ms 202.97.51.121
14 587 ms 583 ms 591 ms 202.97.33.125
15 593 ms 593 ms 592 ms 202.97.43.198
16 596 ms 591 ms 597 ms 58.60.8.22
17 588 ms 591 ms 593 ms 58.60.8.106
18 585 ms 590 ms 592 ms 219.133.40.91

Trace complete.

The best thing about this new link from my perspective is that soon we'll be able to connect to asia.battle.net, great for starcraft, warcraft3, and soon (hopefully) starcraft 2.

It's hosted in korea though, so near to middle term eh?

Anyone care to speculate on whether the coming link to guam will have even better latency than this one?

I must be blind, becuase I can see the packets hop off in Japan, yet the network map doesnt show the Japan POP yet?

FTTH Test Case writes...

I must be blind, becuase I can see the packets hop off in Japan, yet the network map doesnt show the Japan POP yet?

Place a UV light up towards your screen and it will appear on the map. :P

This is better now for connecting to south east asia, but still dodgy with pings of 250ms (acceptable) to 500 ms (unacceptable) during the evenings and on weekends.

I notice both Telstra and Optus get about 110-150ms, as they are going through Perth and then Singapore. Makes churning very tempting even though I would probably regret it in other ways.

Are there any plans (even long term) to be able to connect through Singapore?

Spam eggs sausage writes...

I notice both Telstra and Optus get about 110-150ms, as they are going through Perth and then Singapore.

I'm with Optusnet and traffic to Singapore goes via the US first. Here's a tracert of what to expect if you churned:

Tracing route to singtel.com [165.21.20.180]
over a maximum of 30 hops:

1 2 ms 1 ms 1 ms 165.21.20.180
2 * * * Request timed out.
3 9 ms 19 ms 9 ms riv3-ge0-2.gw.optusnet.com.au [198.142.160.241]
4 11 ms 9 ms 9 ms mas1-ge11-0.gw.optusnet.com.au [211.29.129.41]
5 9 ms 15 ms 11 ms mas3-ge2-0-0-811.gw.optusnet.com.a­ u [211.29.155.11]
6 174 ms 173 ms 174 ms 203.208.148.97
7 177 ms 173 ms 173 ms ge-5-0-0-0.laxow-cr2.ix.singtel.co­ m [203.208.183.157]
8 174 ms 175 ms 172 ms xe-1-0-0-0.laxow-cr1.ix.singtel.co­ m [203.208.183.145]
9 366 ms 389 ms 371 ms so-3-1-0-0.sngc3-cr1.ix.singtel.co­ m [203.208.149.237]
10 377 ms 370 ms 379 ms 203.208.149.22
11 372 ms 364 ms 364 ms GE-0-3-0.westpoint.singnet.com.sg ­ [165.21.12.18]
12 373 ms 373 ms 373 ms 165.21.237.254
13 377 ms 369 ms 375 ms 165.21.20.180

Trace complete.

I think there have been comments in the past that the seamewe (perth to singapore) line is older tech and theres not much capacity to lease, also the extra latency getting to Perth from the east coast makes it less worth while.

The seamewe line is awesome for Perth people though, but I'd be a bit tentative because I'd never know when my isp might decide to route via the east coast to save money, are you in Perth?

By the way, I can get to war3.replays.net but i cant get to w3g.replays.net/doc/cn/2...24218841636.html
This website has had problems in the past on internode due to odd routing issues that arent happening on other isps, is it possible its slipping through the cracks again? Anyone else having wierd problems?

When I use anonymouse.org (a proxy) it works.

So basically war3.replays.net works, but only the front page, deeper links do not work, and this has happened once befor.

Kon-Tiki writes...

but i cant get to w3g.replays.net/doc/cn/2...24218841636.html

Works fine on my node connection here QLD..

The best network config would be a POP at....

wait for it

....THE CENTRE OF THE EARTH.......

How embarrassing, its prolly just me, come to think of it, it was my problem last time too!

Spam eggs sausage writes...

This is better now for connecting to south east asia, but still dodgy with pings of 250ms (acceptable) to 500 ms (unacceptable) during the evenings and on weekends.

South East Asia isn't one coherent region - it's lots of countries and telcos that use very different paths and network qualities. What and where exactly are you trying to get to? (hint: I can't do anything unless I know where it is you want to get to!)

What was that Matthew, Internode is going to put a new link to Euroupe? Yay! :D

Gaz Iggi writes...

I'm with Optusnet and traffic to Singapore goes via the US first. Here's a tracert of what to expect if you churned:

Tracing route to singtel.com [165.21.20.180]

Hrm... thanks for that info.
Its quite different from what I get from einstein.dontexist.net/austrace
when I select to traceroute from optus...

Performing trace to singtel.com...

1 FastEth8-1-0.sb2.optus.net.au (192.168.34.3) 3.384 ms 3.128 ms 2.896 ms
2 203.208.148.205 (203.208.148.205) 109.411 ms 110.155 ms 109.386 ms
3 ge-0-1-4-0.sngtp-dr1.ix.singtel.co­ m (203.208.183.113) 109.505 ms 115.359 ms 119.397 ms
4 ge-1-0-0-0.sngc3-dr1.ix.singtel.co­ m (203.208.173.134) 113.164 ms 112.827 ms 112.729 ms
5 203.208.149.22 (203.208.149.22) 110.334 ms 111.727 ms 113.270 ms
6 GE-0-3-0.westpoint.singnet.com.sg ­ (165.21.12.18) 116.346 ms 114.205 ms 113.109 ms
7 165.21.237.254 (165.21.237.254) 112.176 ms 112.574 ms 111.370 ms
8 * * *
9 * * *

Matthew Moyle-Croft writes...

What and where exactly are you trying to get to?

goin.perfectworld.com.my

Any eta on when we can get a more direct line to asia.battle.net for the awesome blizzard games?

Lots of other isps have had a direct line for ages. Together, lets dash their hubris!

I've noticed that the traceroute to goin.perfectworld.com.my varies depending on the time of day. Both seem to go to Japan, though different IP addresses in japan, however one then goes to malaysia via hong kong, and this is fairly low latency and reliable. (through 203.105.72.153 xe-4-1-0.r20.tokyjp01.jp.bb.gin.nt­ t.net)

The other route seems to go directly to malaysia, and has much higher latency and packet loss. It seems to choose the fast route in the early hours of the morning up till about noon, after which it goes through the poorer route, with pings over 500 ms and significant packet loss. (through 210.171.224.194 AS4788.ix.jpix.ad.jp)

I sent in a ticket with traceroutes showing this puzzling behavior in the hope that perhaps internode can "lock" it into the fast path through 203.105.72.153, but i'm guessing its beyond their control.

Simon Hackett writes...

The new site will be in Tokyo, Japan.

WTF??!!

Here I am on the Gold Coast, unable to get anything over my crappy ADSL 1.5Mb connection, and you guys are too busy putting hardware in... JAPAN??!!

I would kill for an ADSL2+ connection, but noooooo, you guys are to busy providing Internet to the Japanese! What crap is this, are you an Australian or Japanese ISP? One begs to wonder.

Um... isnt everyone else waiting for there exchange to be dslamed? (I have a 8month fault, going on and im not complaining) The Japan POP benefits us all including corp customers.

Ripperjack writes...

Here I am on the Gold Coast, unable to get anything over my crappy ADSL 1.5Mb connection, and you guys are too busy putting hardware in... JAPAN??!!

They not just putting in DSLAMS everywhere, working on making the network better and faster access to sites overseas, imoproving webmail etc, etc..they doing million of things!

Spam eggs sausage writes...

I sent in a ticket with traceroutes showing this puzzling behavior in the hope that perhaps internode can "lock" it into the fast path through 203.105.72.153, but i'm guessing its beyond their control.

Most likely you will find that it's the ISP in Malaysia shifting traffic around at different times of the day to balance or better utilise their links.

Better for them isn't necessarily better for everyone else :)

A manual frob on our network isn't practical and might not even be possible depending on what they are doing at the remote end.

Ripperjack writes...

you guys are to busy providing Internet to the Japanese! What crap is this, are you an Australian or Japanese ISP?

That quote is just gold

Ripperjack writes...

What crap is this, are you an Australian or Japanese ISP? One begs to wonder.

A Point of Presence isn't an "Internet Service Offering". Internode hasn't started selling in Japan, they, rather, are simply provider a better connection to Japan.

One begs, politely, to wonder if you actually read the first post that explains what this was/ is all about. :)

Ripperjack writes...

unable to get anything over my crappy ADSL 1.5Mb connection

What about 8mbit? Or is there something I'm missing?

Ripperjack writes...

you guys are to busy providing Internet to the Japanese! What crap is this, are you an Australian or Japanese ISP? One begs to wonder.

you dont know what you are talking about!

damo81 writes...

you dont know what you are talking about!

Did you notice his quip? ;-)

Also his profile says he is on 512

Devils Advocate writes...

That quote is just gold

Surely that's a troll.... surely...